Role of naive-derived T memory stem cells in T-cell reconstitution following allogeneic transplantation

Abstract

Early T-cell reconstitution following allogeneic transplantation depends on the persistence and function of T cells that are adoptively transferred with the graft. Posttransplant cyclophosphamide (pt-Cy) effectively prevents alloreactive responses from unmanipulated grafts, but its effect on subsequent immune reconstitution remains undetermined. Here, we show that T memory stem cells (TSCM), which demonstrated superior reconstitution capacity in preclinical models, are the most abundant circulating T-cell population in the early days following haploidentical transplantation combined with pt-Cy and precede the expansion of effector cells. Transferred naive, but not TSCM or conventional memory cells preferentially survive cyclophosphamide, thus suggesting that posttransplant TSCM originate from naive precursors. Moreover, donor naive T cells specific for exogenous and self/tumor antigens persist in the host and contribute to peripheral reconstitution by differentiating into effectors. Similarly, pathogen-specific memory T cells generate detectable recall responses, but only in the presence of the cognate antigen. We thus define the cellular basis of T-cell reconstitution following pt-Cy at the antigen-specific level and propose to explore naive-derived TSCM in the clinical setting to overcome immunodeficiency. These trials were registered at www.clinicaltrials.gov as #NCT02049424 and #NCT02049580.

Figures

Figure 1

Pt-Cy preferentially depletes proliferating effector/memory T cells. (A) Representative (out of 10) Ki-67 and HLA-DR expression in PB CD3+ T cells at day 3 and day 7 post–haplo-HSCT. (B) Mean ± standard error of the mean (SEM) frequency (n = 22; each dot represents a patient) of Ki-67+ T cells with a given differentiation phenotype at day 3 post–haplo-HSCT. +P < .05 vs TRTE; *P < .05 vs TN; Wilcoxon test. (C) Mean ± SEM frequency (n = 23) of CD4+ and CD8+ T cells at day 3 and day 7 post–haplo-HSCT. P < .05 vs day 3; Wilcoxon test. (D) CFSE dilution and CD25 expression by TN and TMEM following incubation with auto-APCs or allo-APCs for 3 days. The gate in black identifies CFSElo proliferating (ie, alloreactive) cells, whereas that in gray CFSEhi nonproliferating cells. (E) CD45RO expression by CFSElo and CFSEhi cells, identified as in panel D, originally sorted as TN (top) or TMEM (bottom).

Figure 2

Donor TSCM dominate the peripheral T-cell compartment following pt-Cy. (A) FACS analysis of PB T cells at day 7 post–haplo-HSCT. Donor (D; red) and recipient (R; dark gray) cells are identified by an antibody recognizing the mismatched HLA-A*02. Light-gray cells in the background are T cells from the PB of a healthy donor. (B) Median ± SEM frequency of D and R T cells (identified as in panel A) with a given differentiation phenotype in patients at day 7 post–haplo-HSCT (n = 7, CD4+; n = 6, CD8+). Only donor-recipient pairs whose mismatched HLA could be investigated by flow cytometry are included. *P < .05 vs day 7 D cells; Mann-Whitney test. (C) Percent CFSElo CD8+ T cell subsets from a healthy donor (HD) and a recipient (R) at day 41 post–haplo-HSCT after PBMC culture with 1 (gray, serving as a nonproliferating control) or 50 ng/mL IL-15 (black) for 8 days. N/D, not detected. (D) Mean ± SEM CFSElo CD8+ T-cell subsets (calculated as in panel C) from healthy donors (n = 6) and haplo-HSCT patients (n = 5). TMEM, CD45RO+ memory T cells. *P < .05, Mann-Whitney test. (E) Combinations of IFN-γ, IL-2, and TNF production in gated TSCM from patients (n = 3) and in T-cell subsets from healthy donors (HD; n = 4). *P < .05, permutation test.

Figure 3

Putative TN-cell origin of posttransplant TSCM. (A) Representative frequency (out of 12) of CD45RO−CCR7+ T cells in a patient a day 0 and day 3 post–haplo-HSCT. (B) Representative (out of 12) CD31 and CD95 expression on NL-CD4+ T cells from the PB and BM of a donor and from the PB of the related recipient at different times post–haplo-HSCT. A scheme with the nomenclature of subsets according to phenotype is depicted on the left. NL, CD45RO−CCR7+. Numbers in plots indicate the percentage of cells in the gates. (C) Mean ± SEM CD31 expression on PB CD4+CD95− cells and CD4+ TSCM from marrow donors (D) and CD4+ TSCM from the related recipients (R; n = 12) at day 7 post–haplo-HSCT. *P < .05 vs D TSCM; Wilcoxon test. (D) Fold change in CD95 median fluorescence intensity (MFI) in different T-cell subsets (n = 19) between day 3 and day 7. *P < .05 vs NL; Wilcoxon-paired test. (E) Representative analysis of CD95 expression by TN cells following incubation with different stimuli. Histograms are referred to the CD25−CD69− nonalloreactive population, as specified in the text. (F) Summary of the data obtained in panel E (n = 8; 4 independent experiments; *P < .05 vs allo-APCs, Wilcoxon-paired test). (G) CFSE dilution and CD31 expression by FACS-sorted T-cell subsets following incubation with allo-APCs for 5 days. D, donor; R, recipient.

Figure 4

Persistence and memory differentiation of antigen-specific TN. (A) Frequency of CD8+ T cells in PBMCs from 2 CMV− donors and matched CMV+ recipients at different time points post–haplo-HSCT. (B) Frequency and phenotype of MART-1– and WT1-specific CD8+ T cells identified by MHC class I tetramers. Tetramer+ T cells are overlaid on top of the total CD8+ T-cell population depicted in gray. In panels A and B, numbers indicate the percentage of cells identified by the gates. (C) Frequency of MART-1+ (filled gray circles) and WT1+ (blank circles) CD8+ T cells with a TN-cell phenotype in donors (D) and recipients (R) at day 45 and day 90 post–haplo-HSCT. P < .05, Mann-Whitney test. (D) Mean ± SEM frequency of the cells identified in panel B.

Figure 5

Antigen-specific memory T cells may survive pt-Cy and expand in the host in the presence of the cognate antigen. (A) Frequency of CMV-specific memory T cells in haplo-HSCT recipients (R) at day 45 and day 90 and in related donors (D). (B) Representative analysis of donor-derived CMV-specific T-cell responses detected by simultaneous analysis of the mismatched HLA (in this case the donor was HLA-A*02−) and intracellular IFN-γ following stimulation with CMV pp65 overlapping peptide mix. Plots show Aqua−CD3+ cells. (C-D) TCRB clonal composition of CMV-specific CD4+ (C) and CD8+ (D) T cells in CMV+/+ donor/recipient pairs. Overlapping sequences are highlighted in gray. d, day after haplo-HSCT.

Figure 6

Poor expansion of adoptively transferred memory T cells in the absence of cognate antigen. (A) MHC class I tetramer identification of CD8+ T cells specific for Flu IK9 and Flu GL9 epitopes in the PB of marrow donors and the related recipients (R) at day 90 post–haplo-HSCT. (B) TNF and IFN-γ production by CD4+ and CD8+ T cells from the haplo-HSCT #10 donor/recipient pair (CMV+ and CMV−, respectively) following in vitro stimulation with CMV pp65 peptide pool. In panels A and B, numbers indicate the percentage of cells identified by the gates. (C) Summary of the frequency of CD3+ natural killer T cells binding CD1d/PBS57 tetramer and Flu and CMV-specific CD4+ and CD8+ T cells from the PB of donor (D) and the related recipients (R) at day (d) 45 and d90 post–haplo-HSCT. *P < .05, Mann-Whitney test. (D) Differentiation phenotypes of the transferred antigen-specific T cells identified in panel C. Data are presented relative to total memory T cells.

Figure 7

Proposed model of the cellular mechanisms leading to T-cell reconstitution following haplo-HSCT and pt-Cy. Naive (TN) and memory T cells (TMEM) are infused in the recipient with the BM. Allogeneic antigens (allo-Ags), as well as inflammatory/homeostatic cytokines, the availability of which increases after chemotherapy, induce T-cell activation. Proliferating (HLA-DR+, Ki-67+) T cells uniformly acquire an effector phenotype, irrespective of their original differentiation status, and are preferentially depleted by Cy, given at day 3 and 4 after HSCT. T stem cell memory (TSCM) is the dominant peripheral T-cell subset at day 7, likely originating from TN that survived Cy. In the following weeks, naive-derived TSCM generate memory cells in response to exogenous antigens and, presumably, homeostatic cytokines. Adoptively transferred TMEM, which have survived Cy, expand to detectable levels in the circulation only in the presence of the cognate antigen. Whether T-cell memory can persist in the haplo-HSCT individual in the absence of the cognate antigen is currently unknown.